Postharvest Biology and Technology 15 (1999) 207 225

Quality measurement of fruits and vegetables

Judith A. Abbott
Horticultural Crops Quality Laboratory, Plant Science Institute, Agricultural Research Ser6ice, USDA, 002, Belts6ille, MD 20705, USA Received 30 June 1998; accepted 11 November 1998

Abstract To investigate and control quality, one must be able to measure quality-related attributes. Quality of produce encompasses sensory attributes, nutritive values, chemical constituents, mechanical properties, functional properties and defects. Instrumental measurements are often preferred to sensory evaluations in research and commercial situations because they reduce variations in judgment among individuals and can provide a common language among researchers, industry and consumers. Essentially, electromagnetic (often optical) properties relate to appearance, mechanical properties to texture, and chemical properties to avor (taste and aroma). Instruments can approximate human judgments by imitating the way people test the product or by measuring fundamental properties and combining those mathematically to categorize the quality. Only people can judge quality, but instruments that measure quality-related attributes are vital for research and for inspection. 1999 Elsevier Science B.V. All rights reserved. Keywords: Fruits; Quality measurement; Quality-related attributes; Vegetables

1. Introduction The term quality implies the degree of excellence of a product or its suitability for a particular use. Quality is a human construct comprising many properties or characteristics. Quality of produce encompasses sensory properties (appearance, texture, taste and aroma), nutritive values, chemical constituents, mechanical properties, functional properties and defects.

Shewfelt (1999) points out that quality is often dened from either a product orientation or a consumer orientation. However, I personally have difculty divorcing the two viewpoints and tend to think in terms of instrumental or sensory measurements of quality attributes that combine to provide an estimate of customer acceptability. Of course, one must always remember that there is more than one customer in the marketing chain. The next person or institution in the following

0925-5214/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 5 - 5 2 1 4 ( 9 8 ) 0 0 0 8 6 - 6

208

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

chain can be considered a customer by the previous one: grower, packer, distributor and/or wholesaler, retailer, produce manager, shelf stocker, shopper, and nally the ultimate consumer who actually eats the product. Each passes judgment, and each has its own set of quality or acceptability criteria, often biased by personal expectations and preferences. The component attributes of quality vary with context. The choice of what to measure, how to measure it, and what values are acceptable are determined by the person or institution requiring the measurement, with consideration of the intended use of the product and of the measurement, available technology, economics and often tradition. For grades and standards of a product, the denition of quality is formalized and institutionalized so it has the same meaning for everyone using it. Shewfelt (1999) suggests that the combination of characteristics of the product itself be termed quality and that the consumers perception and response to those characteristics be referred to as acceptability. The dictionary denition of quality encompasses both concepts (Websters; Neufeldt, 1988). People use all of their senses to evaluate quality: sight, smell, taste, touch, and even hearing. The consumer integrates all of those sensory inputs appearance, aroma, avor, hand-feel, mouth-feel and chewing sounds into a nal judgment of the acceptability of that fruit or vegetable. Instrumental measurements are preferred over sensory evaluations for many research and commercial applications because instruments reduce variations among individuals, are more precise, and can provide a common language among researchers, industry and consumers. However, the relationship of the instrumental measurement to sensory attributes (e.g. descriptive analysis) and the relationship of those sensory attributes to consumer acceptability must be considered (see Shewfelt, 1999). Instruments may be designed to imitate human testing methods or may be statistically related to human perceptions and judgments to predict quality categories. Essentially, appearance is detected instrumentally by measuring electromagnetic (usually optical) properties, texture, by mechanical properties, and avor (taste and aroma) by chemical properties.

Some sensors are based on signals not detectable by humans: for example, near infrared, X-ray, magnetic resonance and electrochemical. Fruits and vegetables are notoriously variable, and the quality of individual pieces may differ greatly from the average. It is essential to determine statistically the number of pieces and the number of measurements per piece required to achieve signicant, representative sampling. Sampling predicts the average quality, and perhaps quality distribution, of the lot. Sampling does not identify the undesirable or the outstanding individual fruit or vegetable. Sorting is necessary to segregate the undesirable, acceptable and outstanding individual pieces and to ensure the uniform quality required commercially. Sorting requires high-speed, non-destructive sensors to measure several attributes on each piece of fruit or vegetable, a means to combine those measurements into a classication decision, and a mechanism to physically place the piece into its proper category. Often empirical methods developed to measure some particular quality attribute actually measure ripeness. Physiological processes involved in ripening and senescence occur together in a fairly predictable pattern. For example, measuring color might appear to be adequate to estimate rmness in peaches and tomatoes at harvest. However, growing conditions or postharvest treatments may decouple the physiological processes such that indirect measurements no longer predict quality as expectede.g. the fruit softens but the color does not change. So it is essential to recognize what is really being measured and to respect the limitations of indirect measurement. It is usually assumed that there is a consistent and unidirectional, although not always linear, relationship between the sensor response and the amount of the constituent or attribute. This is generally true of quantitative measurements, such as measuring pigment content by light absorbance or hardness by force to achieve a given deformation. However, this unidirectional relationship is not necessarily the case for acceptability. Consumers often prefer intermediate levels of a particular attribute to either very low or high levels; for instance, a rm tomato is preferred to one that is too hard or too soft. Some attributes are bino-

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

209

mial; for example, absence of decay is acceptable, presence is unacceptable. Research continues to establish the relationships among instrumental measurements, sensory intensity responses, and perceived quality or acceptability. Methods to measure quality and quality-related attributes have been developed over centuries, with instrumentation appearing in about the past 80 years. Most recently, the emphasis has been on developing sensors for real-time, non-destructive sorting. There have been numerous reviews of technologies for non-destructive quality measurement, including those of Gunasekaran et al. (1985), Chen and Sun (1991), Tollner et al. (1993), Brown and Sarig (1994), Chen (1996), NRAES (1997), Dull et al. (1996) and Abbott et al. (1997). Some instrumental measurements both destructive and non-destructive that are extensively used in horticulture and several emerging technologies will be reviewed briey. Due to space limits, only seminal or representative references are given.

2. Electromagnetic technologies The electromagnetic spectrum encompasses, from longest to shortest wavelengths, radiowave, microwave, ultraviolet, visible light, infrared, Xray and gamma-ray radiation. Optical properties indicate the response of matter to visible light wavelengths (400700 nm, sometimes given as 380 770 nm), and usage is often extended to include ultraviolet (UV, 4 400 nm) and near infrared (NIR, 700 or 770 to 2500 nm). For convenience, the term light will be used loosely to encompass the UV, visible and NIR ranges in the following discussion unless otherwise qualied. However, now the UV is seldom used for horticultural quality measurement. X-ray is discussed separately. The other wavelength ranges have not been successfully applied to quality measurement of horticultural commodities.

from the product carries information used by inspectors and consumers to judge several aspects of quality; however, human vision is limited to a small region of the spectrum. Colorimeters measure light in terms of a tristimulus color space that relates to human vision; they are restricted to the visible light region. Some quality features respond to wavelengths in regions outside the visible spectrum. Spectrometers and spectrophotometers measure wavelengths in the UV, visible and NIR spectral regions; instruments are optimized for a particular wavelength range. Optical properties are based on reectance, transmittance, absorbance, or scatter of light by the product. When a fruit or vegetable is exposed to light, about 4% of the incident light is reected at the outer surface, causing specular reectance or gloss, and the remaining 96% of incident energy is transmitted through the surface into the cellular structure of the product where it is scattered by the small interfaces within the tissue or absorbed by cellular constituents (Birth, 1976) (Fig. 1). The complex physical structure of tissues creates an optically dense product that is difcult to penetrate and alters the pathlength traveled by the light so that the amount of tissue interrogated is not known with certainty. Most light energy penetrates only a very short distance and exits near the point of entry; this is the basis for color. But some penetrates deeper (usually a few millimeters, depending on optical density) into the tissues and is altered by differential absorbance of

2.1. Optical properties

Appearance is a primary factor in quality judgments of fruits and vegetables. Light reected

Fig. 1. Incident light on a fruit or vegetable results in specular reectance (gloss), diffuse reectance from features at depths to about 5 mm (body relectance or interactance), diffuse transmittance, or absorbance. Color results from very shallow diffuse reectance.

210

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

various wavelengths before exiting and therefore contains useful chemometric information (Fig. 1). Such light may be called diffuse reectance, body reectance, diffuse transmittance, body transmittance, or interactance; these terms are not always clearly distinguished. Color is the basis for sorting many products into commercial grades, but concentration of pigments or other specic constituents might provide a better quality index (K.H. Norris, personal communication, 1967; Lancaster et al., 1997). Color relates more directly to consumer perception of appearance, pigment concentration may be more directly related to maturity, and concentration of certain other constituents relates more closely to avor. Color of an object can be described by several color coordinate systems (Clydesdale, 1978; Francis, 1980; Hunter and Harold, 1987; Minolta, 1994). Some of the most popular systems are RGB (red, green and blue), which is used in color video monitors, Hunter L a b, CIE (Commission Internationale de lEclairage) L* a* b*, CIE XYZ, CIE L* u* v*, CIE Yxy, and CIE LCH. These differ in the symmetry of the color space and in the coordinate system used to dene points within that space. Of greatest importance to instrumental measurement are the tristimulus methods of the CIE and the similar Hunter system. According to CIE concepts, the human eye has three color receptors red, green and blue and all colors are combinations of those. The most commonly used notations are the CIE Yxy color space devised in 1931, the Hunter L a b developed in 1948 for photoelectric measurement, and the CIE L* a* b* color space (Fig. 2) devised in 1976 to provide more uniform color differences in relation to human perception of differences. Color is measured by colorimeters in which the sensors are ltered to respond similarly to the human eye. Most provide automatic conversion among several color coordinate systems. Spectrophotometric data can be multiplied by three standard spectra representing the responses of the average human eyes three color receptors and then tristimulus values can be calculated. Automated color sorting is commercially used on packing lines for apples, peaches and several other horticultural commodities.

Fig. 2. CIE L*a*b* color space. L* indicates lightness, from white =100 to black = 0; a* and b* are XY color coordinates indicating color directions; a* is the red green axis, b* is the yellow and blue axis; the center is achromatic gray. Hue is location around the circumference and saturation is distance from center. Adapted from Minolta (1994) with permission.

Chemical bonds absorb light energy at specic wavelengths, so some compositional information can be determined from spectra measured by spectrophotometers or spectrometers. Within the visible wavelength range, the major absorbers are the pigments: chlorophylls, carotenoids, anthocyanins and other colored compounds. Water, carbohydrates, fats and proteins have absorption bands in the NIR region. Williams and Norris (1987) list some of the major absorption wavelengths for pigments, fats, proteins, carbohydrates and water. There is a vast literature on optical measurement of pigments and other constituents and on their relationships to maturity and quality. Currently multiwavelength or whole-spectra analytical methods are being developed for non-destructive determination of soluble solids, acids, starches and ripeness (Fig. 3). Starch or soluble solids (SS) content can be determined in intact fruit (apple, citrus, kiwifruit, mango, melons, onion, peach, potato and tomato) with R 2 : 0.93

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

211

and SEC (standard error of calibration): 0.5% SS (Dull, 1984; Birth et al., 1985; Dull et al., 1989; Kawano et al., 1992, 1993; Murakami et al., 1994; Katayama et al., 1996; Slaughter et al., 1996; Lee et al., 1997; Pieris et al., 1998). Oil content is important in seeds, nuts, and avocados and can be determined using NIR (D.C. Slaughter, personal communication, 1996). Multi- or hyperspectral cameras permit rapid acquisition of images at many wavelengths. Imaging at fewer than ten wavelengths is generally termed multispectral, and more than ten termed hyperspectral. The resulting dataset can be visualized as a cube with the X and Y dimensions being the length and width of the image (in pixels) and the Z dimension being spectral wavelengths; each datapoint is an intensity value. Alternatively, the dataset could be envisioned as a stack of singlewavelength pictures of the object, with as many pictures as the number of wavelengths used. Such imaging provides information about the spatial distribution of constituents (pigments, sugars, moisture, etc.) near the products surface. Differences between sound and damaged tissues in visible and NIR diffuse reectance are useful for detecting bruises, chilling injury, scald, decay lesions and numerous other defects. Bruises on apples and peaches can be detected at specic NIR wavelengths; however, the wavelengths chosen for apple differ between fresh and aged bruises because of drying of the injured tissues

(Upchurch et al., 1994). Discriminant analysis of images of 18 common defects on several apple cultivars at 58 wavelengths between 460 and 1030 nm (Fig. 4) revealed that four wavelengths generally separated sound and damaged tissues: 540, 750, 970 and 1030 nm (Aneshansley et al., 1997; Throop and Aneshansley, 1997). Different wavelengths may be optimal for other commodities (Fig. 5). Differences in images taken at specic wavelengthsmultispectral or hyperspectral imagingand computerized image processing techniques are being used to automate detection and classication of many defects on-line. Such sorters are presently in the advanced commercial testing stage. In addition to imaging technologies, the major advances in spectral analysis in recent years have been in statistical methods. Early analyses used multiple linear regression of raw, rst difference, or second difference spectra (Hruschka, 1987). Later methods used various forms of data reduction such as principal component or partial least squares coupled with multiple regression. Present investigations focus on articial neural networks (Chen et al., 1995; Song et al., 1995) and wavelets for data reduction. Each approach has advantages and disadvantages. Rapid scanning spectrophotometers are available and permit use of all or large parts of the spectrum. Optical-lter instruments or multispectral cameras require wavelength selection, rather than full-spectrum scanning. Limiting

Fig. 3. Visible and near infrared interactance spectrum of Delicious apple. Strongest absorption wavelengths of water, starch and sugar are indicated on the X axis.

212

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

Fig. 4. Multispectral relectance images of apple defects taken with a 35-nm bandwidth at each nominal wavelength. Top: bruise on Delicious after 24 h (from left: standardized reference image at 540 nm, 1010 nm, binarized difference). Bottom: decay on Crispin (from left: 1010 nm, 760 nm, binarized difference. (Aneshansley and Throop, unpublished.)

the number of wavelengths required reduces measurement time, even with acousto-optical or liquid crystal tunable lters, enabling application of optical measurement on-line for sorting operations at commercially acceptable speeds. Of course, limiting the number of wavelengths also reduces computational time; but it may also reduce the chemometric content of the data. Data processing of hyperspectral images is particularly complex, requiring the development of hybrid analyses using both spectroscopic and imaging concepts. New statistical methods are being developed to utilize hyperspectral images efciently for quality assessment. Regardless of the statistical methods, it is critical that the underlying relationship between the measurement and the quality attribute be valid and robust. There must be a fundamental relationship between the wavelength selection and the chemical(s) being sensed or the measurement will ultimately fail.

2.2. Fluorescence and delayed light emission

Fluorescence results from excitation of a molecule by high energy light (short wavelength) and its subsequent instantaneous relaxation with

the emission of lower energy light (longer wavelength). Many agricultural materials uoresce; however, nearly all horticultural applications of uorescence involve chlorophyll. In this review, uorescence refers specically to chlorophyll uorescence unless otherwise stated. Delayed light emission (DLE) is a related phenomenon in which chlorophyll is excited by back reactions of photosynthetic intermediates and then relaxes as in uorescence. Note that uorescence and DLE are light emitted by excited chlorophyll, not merely reected or transmitted light. Peak excitations of chlorophyll are induced by wavelengths around 420 nm (blue) or around 680 nm (red). The peak emission occurs at 690 nm with a small peak between 720 and 750 nm. Chlorophyll uorescence in leaves has a lifetime of about 0.7 ns at 25C (Butler and Norris, 1963); however, it can be detected during continuous illumination by properly ltered detectors and displays characteristic reaction kinetics (Fig. 6). DLE is detectable only in the dark following excitation. Because of photosynthetic kinetics, reproducible measurements of uorescence or DLE are obtained only when excitation is preceded by a dark period, typically 10 min or longer.

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

213

Fig. 5. Multispectral reectance images of tomato with viral ring spot (from left: 540 nm, 700 nm, binarized difference). (Abbott, unpublished.)

Fluorescence measurements of chlorophyll-containing tissue are routinely used for investigations of photosynthetic activity in plant leaves (Schreiber et al., 1975). Chlorophyll content and its photosynthetic capacity are often related to maturity of plant organs and to certain defects or injuries. Fluorescence and DLE have been studied as possible methods for evaluating maturity in fruits and vegetables that lose chlorophyll as they ripen or mature (Jacob et al., 1965; J.N. Yeatman, personal communication, 1967; Nakaji et al., 1978; Chuma and Nakaji, 1979; Chuma et al., 1980, 1982; Forbus et al., 1987, 1992; Abbott, 1996). Physiological stresses that affect chloroplasts or photosynthesis, such as temperature, salinity, moisture, atmospheric pollutants and mechanical damage can also affect uorescence and DLE (Melcarek and Brown, 1977; Smillie and Nott, 1979; Abbott and Massie, 1985; Smillie et al., 1987; Chan and Forbus, 1988; Abbott et al., 1991, 1994, 1997; Toivonen, 1992; Lurie et al., 1994; DeEll et al., 1996; Tian et al., 1996; Woolf and Laing, 1996; Beaudry et al., 1997; Mir et al., 1998). It should be noted in this context that physiological stresses could affect maturity measurements using these principles and vice versa (Abbott et al., 1994, 1997). A relatively new chlorophyll uorescence technique is pulse amplitude modulated (PAM) uorometry which measures features related to quenching due to electron transport, proton pumping of ATPase and pH gradients in the

thylakoid membrane. Studies using PAM uorescence have been reported for following development of injury due to chilling (Lurie et al., 1994) and hot water treatments (Tian et al., 1996; Woolf and Laing, 1996). Woolf and Laing (1996) concluded that uorescence reects the effect of heat on the photosynthetic system in avocado fruit but does not necessarily indicate the overall condition of the skin.

Fig. 6. Fluorescence image of mature green tomato with incipient chilling injury. Lower right half: steady-state uorescence (10 min dark, then 3 min excitation). Upper left half: dark-adapted 13 min, immediate uorescence upon excitation. Dark spots are insect injuries or blossom scar. Bright spots may indicate chilling stress (Abbott, Kim, McMurtrey and Chappell, unpublished).

214

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

Fluorescence or DLE imaging should enable visualization of the spatial distribution of stress responses before visible symptoms develop (Abbott, 1996). This should be of considerable interest in physiological studies of chilling injury and similar stress responses on fruits and vegetables. A uorescence application not based on chlorophyll was the detection of mechanical injury of citrus rind based on uorescence of oils that leaked from damaged oil cells (Uozumi et al., 1987). Seiden et al. (1996) were able to distinguish two apple cultivars and to predict soluble solids content by the uorescence of pasteurized apple juice excited at 265 and 315 nm and measured from 275 to 560 nm (not chlorophyll-related). Seiden et al. point out that sugars do not uoresce but apparently develop in parallel with other compounds that do uoresce.

2.3. X-ray
X-ray has been explored for inspecting the interior of agricultural commodities. The intensity of energy exiting the product is dependent upon the incident energy, absorption coefcient, density of the product and sample thickness. Due to the high moisture content in fruits and vegetables, water dominates X-ray absorption. Anatomical and physiological changes within the tissue of fruits and vegetables such as cell breakdown, water distribution and binding, decay and insect infestation have negative effects on quality. Internal disorders cited in grade standards that should be detectable by X-ray include: cork spot, bitter pit, watercore and brown core for apple; blossom end decline, membranous stain, black rot, seed germination and freeze damage for citrus; and hollow heart, bruises and perhaps black heart for potato. Changes in the density of the internal tissue or distribution of water are usually associated with these defects, chilling injury and insect bodies or feeding. Three X-ray techniques have been used for measuring the interior quality of fruits and vegetables. The most common is two-dimensional radiography, similar to that ordinarily used in medicine. Linescan radiography (Fig. 7), in which the product passes through a vertical plane of

X-ray energy, is most adaptable to on-line measurements. The other method is X-ray computed tomography (CT), also used in medicine. A CT scanner measures X-ray beams over several nonparallel paths through the object, computes a three-dimensional projection, and then computes a slice of that projection. The intensity of the X-ray beam at the detector in all three methods is an integration (or summation) of the energy absorbed along the path from the emitter to detector. The pathlength through tissue increases from the edge toward the center of spherical or cylindrical products, so the apparent intensity is lowest at the center and must be compensated during image processing. An early, successful application of X-ray was the detection of hollow heart in potato (Nylund and Lutz, 1950) (now commercially routine). Another commercially successful application took advantage of increasing density of lettuce heads during maturation (Lenker and Adrian, 1971; Schatzki et al., 1981). Researchers have investigated X-ray techniques for detecting watercore in apple, split pit in peach, the presence of pits in processed olive and cherry, freeze damage in citrus, bruises, and the presence or feeding of insects on numerous commodities (Tollner et al., 1992; Keagy et al., 1996; Schatzki et al., 1997). Brecht et al. (1991) used CT to determine maturity of green tomatoes. With limited success, internal sprouting and ring separations due to microbial rot in onion were detected with X-ray linescan (Fig. 7) (Tollner et al. 1995).

2.4. Magnetic resonance and magnetic resonance imaging (MRI)

Certain nuclei, including 1H, 13C, 31P and 23Na, have a magnetic moment and align along a strong static magnetic eld (Faust et al., 1997). The 1H magnetic resonance is of greatest horticultural interest. A weak pulse of the proper radiofrequency (RF, based on magnetic eld strength) will cause the net magnetic moment to rotate 90. When the RF signal is removed, the moment loses energy and relaxes back to its previous position. Energy released by relaxation induces an RF signal in a receiver coil. Energy loss is differential

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

215

Fig. 7. X-ray images. Top: linescan image for detecting ring separations in onions (from left: original image, Sobel lter processed image); from Tollner et al. (1995). Bottom: computed X-ray tomography image of apples showing bruises and watercore; from Tollner et al. (1992).

and is based on the environment surrounding each nucleus. T1 (spin lattice) relaxation times represent energy loss to the surrounding environment (called the lattice), whereas T2 relaxation times (spinspin) represent loss due to interaction of the spins of multiple nuclei with respect to each other. Images (MRI) are created by applying a magnetic gradient in one direction (phase encoding) and an RF gradient in the other direction (frequency encoding); gradients permit the reconstruction of spatial information. T2 values are often used to describe the biological state of tissues (Faust et al., 1997) and are generally interpreted as the ratio of bound water to free water.

Of particular interest in horticultural applications, areas of greater free water content are brighter than surrounding tissues in MRI, so that disorders involving water distributionwatercore, core breakdown, chilling injury, bruising, decay, presence or feeding of insects, etc.can be visualized. Applications of MR and MRI in horticulture have recently been reviewed by Clark et al. (1997), Faust et al. (1997) and Abbott et al. (1997). Hinshaw et al. (1979) published MRI of Satsuma orange showing the membranes (0.5 mm thick) separating segments. Wang et al. (1988) demonstrated the presence of watercore in an

216

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

apple, as well as showing internal structure. MRI has been used to show morphology, ripening, core breakdown, seeds or pits, voids, pathogen invasion, worm damage, bruises, dry regions, and changes due to ripening, heat, chilling and freezing (Chen et al., 1989; Saltveit, 1991; MacFall and Johnson, 1994; Akimoto et al., 1995; Clark et al.,

1997; Faust et al., 1997). Fig. 8 shows MRI of apples, displaying internal structure and bruising. Another approach to utilizing 1H MR involves interrogating a region within the fruit or vegetable, rather than imaging. Although this method does not provide spatial information and thus limits the information that can be obtained, it

Fig. 8. Magnetic resonance images of intact Delicious apple with three bruises: 1-h-old bruise at bottom, 48-h bruise at top, and small old bruise at upper left. From left: 512 512 pixel resolution, 40 ms echo delay (TE) and 2-mm computed slice; 128 128 pixel resolution, 40 ms TE and 2-mm slice; visible light photograph of the same apple cut through the scanning plane. From Chen et al. (1989).

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

217

dramatically reduces the complexity and cost of instrumentation. Akimoto (1984) patented an MR method for grading fruit based on sugars and/or organic acids. MR has been used to detect pits in halved red sour cherries and correlated to rmness, dry matter, soluble solids content, total acidity and Brix acid ratios in several fruits (Chen et al., 1996; Cho et al., 1993; Stroshine et al., 1994). Currently, MR and MRI are not practical for routine quality testing. MR equipment is expensive and difcult to operate; but, like all technologies, it is becoming cheaper, faster and more feasible for research and specialized applications. MR techniques have great potential for evaluating the internal quality of fruits and vegetables.

3. Mechanical technologies Mechanical properties relate to texture. Harker et al. (1997) recently reviewed the cellular basis of fruit texture and the human physiology involved in its perception. Mechanical tests of texture include the familiar puncture, compression and shear tests, as well as creep, impact, sonic and ultrasonic methods, recently reviewed by Brown and Sarig (1994), Chen (1996) and Abbott et al. (1997). Under mechanical loading, fruits and vegetables exhibit viscoelastic behavior which depends on both the amount of force applied and the rate of loading. However, for practical purposes, they are often assumed to be elastic and loading rate is largely ignored. Measurement of elastic properties requires consideration of only force and deformation, whereas viscoelastic measurement involves functions of force, deformation and time. Nonetheless, because even the rmest fruits and vegetables do have a viscous component to their force/deformation (F/D) behavior, loading rate (test speed) should be held constant in instrumental tests and should be reported. The viscous component has minimal contribution to perceived texture in most rm fruits and vegetables (e.g. apple or carrot), but is quite signicant in soft fruits, notably tomato, cherry and citrus, particularly contributing to hand-feel. That is why a creep or relaxation measurement is often

more suitable for the latter products than is a puncture test. The viscous component of texture is important in determining bruise resistance, even in rm products like potato (Bajema et al., 1998). Although bruise resistance (as distinguished from the presence of bruises) is not considered in evaluating the quality of individual specimens, it should be considered in evaluating the quality of breeding lines in commodities where bruising causes commercial loss. The viscous component is also important in developing handling equipment. Most non-destructive mechanical methods measure elastic properties (e.g. modulus of elasticityYoungs modulus) at very small deformations. Modulus of elasticity measures the capacity of the material to take elastic deformation and is the stressstrain ratio, commonly measured by the slope of the F/D curve prior to rupture for a tissue specimen with constant cross-sectional area (Fig. 9). (Stress is force per unit area; strain is deformation as proportion of initial length.) It should be noted that fundamental tests and material properties measurements like the elastic modulus were developed by materials engineers to study the strength of materials for construction or manufacture. Once the failure point of such a material is exceeded, the material is of little interest. Thus materials engineers would not be interested in the portion of the F/D curve beyond the bioyield point, certainly not beyond the rupture point. The food scientist, on the other hand, is interested in the breakdown of the food in the mouth until it is swallowed. Does peach esh melt or is it stringy? Does apple esh break crisply or is it mealy? Is a baked potato waxy or mealy? As Bourne (1982) pointed out, food texture measurement might be considered more as a study of the weakness of materials rather than strength of materials. In fact, both strength and breakdown characteristics are important components in the texture of fruits and vegetables.

3.1. Quasi static force/deformation tests

Puncture or compression tests made at relatively low speeds, typical of such instruments as the MagnessTaylor fruit rmness tester and electronic universal (F/D) testing instruments, are

218

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

Fig. 9. Force/deformation (F/D) curves for cylindrical tissue specimens (10 mm high 15 mm diameter) from rm and soft apples under constant strain-rate (0.42 mm s 1) compression. F/D curves for puncture are similar except that they are usually terminated at the maximum deformation.

considered quasi static. Typical F/D curves for cylindrical apple specimens under constant strainrate compression are shown in Fig. 9. The portion of the initial slope up to inection represents non-destructive elastic deformation. Beyond that portion, cells start to rupture and there may be a bioyield point where a noticeable change in slope occurs before the rupture point at which signicant tissue failure occurs. In some F/D curves, including those in Fig. 9, bioyield may not be distinguishable from rupture (Bourne, 1965). Beyond rupture, the force may again increase, level off, or decrease as deformation increases. At the maximum deformation point, the probe is withdrawn and the force diminishes until contact is lost. Puncture F/D curves appear similar to compression curves. Firmness of horticultural products can be measured by compression or puncture with various probes at different force or deformation levels, depending on the purpose of the measurement and how the quality attributes are dened. Horticulturists tend to dene rmness as the maximum force attained, although sometimes the rupture or bioyield force is used. The maximum force, re-

gardless of where it occurs, is dened as rmness in the popular MagnessTaylor penetrometer test and usually also in the Kramer multiblade shear test (widely used by the food processing industry to test fruits and vegetables). On the other hand, the slope of the F/D curve, reecting elastic modulus, is often used by materials engineers as an index of rmness. Because sample dimensions, and thus stress and strain, are seldom known precisely in most food tests, this slope measurement should be termed apparent elastic modulus (but seldom is). Elastic modulus can be measured non-destructively, whereas bioyield and rupture by denition require some cellular damage. The best relationships to sensory rmness, hardness and crispness are obtained with forces at or beyond deformations that cause tissue damage (Mohsenin, 1977; Bourne, 1982). Therefore a nondestructive measurement is unlikely to produce excellent prediction of these textural attributes or of MagnessTaylor (or similar) test values, although useful levels of prediction may be attained in tissues where elastic modulus and rupture strength are closely correlated. Quasi static tests do not predict impact properties (G.M. Hyde, 1998, personal communcation).

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

219

Penetrometer testers such as the Magness Taylor (MT) Fruit Firmness Tester (often improperly called a pressure tester), the Effe-gi, McCormick and Lake City Electronic Pressure Tester (all derived from the MT), and the similar U.C. Firmness Tester, are widely used for rm-to-hard fruits and vegetables. The MT tester was developed primarily as an objective measurement of picking maturity (Magness and Taylor, 1925), not as a postharvest quality measurement. Haller (1941) reviewed the use of the MT and its suitability for maturity and storage tests of a number of commodities. A research committee proposed a standardized method for making MT measurements (Blanpied et al., 1978) that also should be applied to related testers. Penetrometer measurements are moderately well correlated with human perception of rmness and with storage life, and consequently this technique has received widespread acceptance for a number of horticultural commodities, such as apple, cucumber, kiwifruit, pear and peach. However, several studies have shown that caution is warranted in the use of such tests for determining quality. Bourne (1979) stated that, while MT is almost the only method used by horticulturists to measure texture, there needs to be a wider understanding of the multidimensional nature of texture and the fact that rmness is only one of the group of properties that constitute texture. Two decades later, the need for this understanding persists. Soft juicy fruits such as tomato, cherry, citrus and various berry fruits often have a signicant viscous component to their texture. Instruments that measure creep (deformation under a constant load for a specic time) are popular for measuring rmness of soft horticultural commodities. There is currently no commonly accepted commercial instrument for this measurement. Although the Cornell rmness tester (Hamson, 1952) is sometimes used to measure tomato rmness, the loads and times applied are not standardized. Another related measurement that is sometimes used is relaxation (decrease in force with time at a constant deformation). There also is no commonly accepted standard method for this measurement.

Plunger tip geometry (i.e. size and shape) must be carefully considered in any mechanical test. For example, Jackman et al. (1990) found that a at-plate probe could not detect rmness differences between sound tomatoes and those with slight chilling injury, whereas a rounded punch probe did. The common geometries for probes include spherical, rounded or at-surfaced cylinders (diameter much smaller than that of the specimen), cones and at plates (diameter greater than diameter of contact with the specimen). The instrument, probe geometry, deformation or penetration distance and rate of loading (speed) used in texture tests should always be reported. A recent problem with rmness tests is the use of probes of varied geometries under the generic term MagnessTaylor. The authentic MT probesthere are twohave gently

Fig. 10. Geometry and dimensions of popular puncture probes. Left: authentic Magness Taylor (MT) probes (dimensions shown in inches for precision because originial MT specications are in inches; 7/16 inch :11.1 mm, 5/16 inch : 7.9 mm). Right: probes with the same diameter but hemispherical tip. Simple cylindrical probes and other geometries are also used for puncture tests.

220

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

rounded tips (of specied radii of curvature), not hemispherical or at tips (Fig. 10) (J.F. Cook, Jr., Past President, D. Ballauff Mfr. Co., Laurel, MD, personal communication, 1998). Because of the curvature of the MT probes and the fact that rmness as measured in puncture is a combination of shear and compression, it is not possible to convert measurements made with one MT probe to the other MT size, or to convert to or from values for probes of other geometries (Bourne, 1982). Due to their low speeds and often destructive nature, compression and penetration F/D techniques are not very adaptable for on-line sorting of horticultural products. Numerous methods for rapid, non-destructive texture measurement have been developed based on responses to very small deformations, vibrations, air puffs (Prussia et al., 1994) and impacts. These were recently reviewed by Abbott et al. (1997). None has gained widespread acceptance, nor has any been shown to predict MT rmness reliably, although correlations as high as 0.94 have been obtained in some tests (Abbott and Massie, 1998).

3.3. Sonic and ultrasonic 6ibration

Sonic (or acoustic) vibrations encompass the audible frequencies between about 20 Hz and : 15 kHz; ultrasonic vibrations are above the audible frequency range ( \ 20 kHz). Sonic and ultrasonic waves can be transmitted, reected, refracted or diffracted as they interact with the material. Wave propagation velocity, attenuation and reection are the important parameters used to evaluate the tissue properties of horticultural commodities. When an object is excited at sonic frequencies, it vibrates. At particular frequencies it will vibrate more vigorously, causing amplitude peaks; such a condition is referred to as resonance. Resonant frequencies are related to elasticity, internal friction or damping, shape, size and density. The rmer the esh, the higher the resonant frequency for products of the same size and shape. The traditional watermelon ripeness test is based on the acoustic principle, where one thumps the melon and listens to the pitch (frequency) of the resonance. The sonic vibration method is truly non-destructive and is suitable for rapid rmness measurement. Sonic measurement generally represents the mechanical properties of the entire fruit, unlike puncture or compression which samples localized tissues. Stiffness coefcients incorporating resonant frequency and mass can compensate for size differences; however, shape also inuences sonic rmness measurements (Lu and Abbott, 1997). Sonic measurement is an excellent means for following changes in individuals over time in research applications and is suitable for determining average rmness of lots of fruit, but has not always proved capable of predicting the MT rmness of individual fruit (Abbott et al., 1968, 1995; Saltveit et al., 1985; Armstrong et al., 1990; Peleg et al., 1990; Stone et al., 1994; Shmulevich et al., 1995). In apples, correlation coefcients between sonic stiffness coefcients and MT values have ranged from 0.5 to 0.9, depending on cultivar, storage conditions, MT range, sample size and method of excitation. Correlations of \0.9 have been obtained for kiwifruit (Abbott and Massie, 1998). Abbott et al. (1997) reviewed various appli-

3.2. Impact
There are various forms of impact test in which the force/time or force/frequency spectrum is recorded as the fruit is dropped onto a force transducer or as the transducer impacts the fruit. A number of impact parameters have been related to rmness: peak force, ratio of peak force to time-to-peak (or time squared), coefcient of restitution, contact time and frequency spectrum. Bajema et al. (1998) developed an impact tester for research on bruise resistance. Delwiche et al. (1989) developed a single-lane rmness sorting system with a rate of 5 fruit s 1 for sorting pear and peach. Delwiche and Sarig (1991) and Sugiyama et al. (1994) developed probe impact sensors that use peak force or coefcient of restitution for rmness measurement. McGlone and Schaare (1993) and Patel et al. (1993) reported on impact rmness instruments for research, quality control and sorting. Zapp et al. (1990) developed an instrumented sphere, simulating a fruit, to determine the impact history during handling, packing and transport of fruits and vegetables.

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

221

cations, methods and theoretical studies of this method. Ultrasonic waves can be transmitted, reected, refracted or diffracted as they interact with the material. Wave propagation velocity, attenuation and reection are the important ultrasonic parameters used to evaluate the tissue properties of horticultural commodities. However, because of the structure and air spaces in fruits and vegetables, it is difcult to transmit sufcient ultrasonic energy through them to obtain useful measurements. Upchurch et al. (1987) attempted to detect bruises in apple, and Galili et al. (1993) found that ultrasonic measurements could be used for rmness determination in some fruits but that a more powerful ultrasonic source is required to penetrate others. Despite numerous studies, few applications have developed (Abbott et al., 1997).

4. Electrochemical technologies The concentration of volatiles within a fruit or vegetable increases as it ripens and their release to the surrounding atmosphere is responsible for the products pleasing aroma. Aromatic and non-aromatic volatiles are released, including ethylene, ethyl esters, acetaldehyde, ethanol and acetate esters. The electrical conductivity of semiconductor gas detectors, based on different polymers and metal oxides, decreases on exposure to volatiles. Although the detectors are not specic for particular volatiles, each type is generally sensitive to a particular class of compounds. A battery of several detectors can produce a ngerprint that may indicate maturity or presence of some disorders. The electronic sniffer concept (Benady et al., 1995; Simon et al., 1996) has been tested on apples, blueberries, melons and strawberries. Further research is needed to explore the selection of semiconductors and to relate the ngerprints to quality categories.

5. Statistical methods Signicant advances have been made in the practical application of classical statistical meth-

ods and in the development of new methods for relating instrumental data to quality assessments, quality categories and acceptability judgments. Statistical methods are used for data reduction the selection of measurement variables, such as wavelength, for predicting qualityand for product classication. Some of the newer methods in this context are partial least squares, principal components, factor analyses, articial neural networks and wavelet analysis. Applications of numerical modeling are aiding in the understanding of sensor responses, such as nite element modeling of vibrational behavior or modeling progressive changes in chemical composition. Advances in machine vision and image processing require statistical pattern recognition, involving histogram analysis, edge recognition, dynamic thresholding and visual texture analysis. There is growing recognition that quality is a multifaceted attribute and there is increasing interest in sensor fusion or combining several inputsdifferent measurements from a single sensor, measurements on different parts of a product (views), or measurements from different sensorsinto a quality classication decision (Chen et al., 1995; Heinemann et al., 1995; Ozer et al., 1995). In addition to classical discriminant, cluster and principal component analyses, researchers are investigating newer computer-intensive multivariate statistical methods such as recursive classication trees and articial neural networks. However, it is essential to use judgment in the application of these statistical methods. In the absence of fundamental relationships between the measurement and the quality attributefor example, light energy absorbance by chemical bonds in spectrophotometric measurementspurious relationships may be found under particular circumstances. When the circumstances change, the relationship is no longer there and the new measurement no longer predicts the quality attribute. Much time and money can be invested in a meaningless measurement if care is not taken to ensure that it is fundamentally valid and robust. One cannot substitute statistical analyses for the knowledge and judgment of the investigator.

222

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225

6. Overview and conclusions Quality is not a single, well-dened attribute but comprises many properties or characteristics. Statistical combination of measurements by several sensors will increase the likelihood of predicting overall quality. However, sensor testing and calibration must include a wide range of conditions. It is important that what is really being detected is understood so the limitations are appreciated. Of course, there are different requirements for laboratory and industry applications. Appearance is one of the major factors the consumer uses to evaluate the quality of fruits and vegetables, and measurement of optical properties has been one of the most successful instrument techniques for assessing quality. Many products are routinely sorted for color. Optical methods are being developed for on-line detection of surface defects based on optical measurements in the visible or NIR regions. Optical systems, especially in the NIR region, and newer software make it possible to detect carbohydrates, proteins and fats that may improve quality indexes. It is likely that on-line NIR sensing of soluble solids will be routine in the near future. Multispectral and hyperspectral imaging provide spectral information at multiple wavelengths in addition to spatial information. Differential reectance of various wavelengths from sound and defective tissue enable detection and often identication of the defects. Fluorescence can detect surface damage on products with signicant amounts of chlorophyll; laboratory instruments are readily available. Electronic sniffers based on the responses of semiconducting materials to volatiles may be able to accurately classify a number of fruits into ripeness or aroma quality categories. X-ray inspection systems are now used to detect internal defects on-line in some limited applications, but the increasing sensitivity of the equipment and the development of rapid image processing could soon make this technology more available. MRI has great potential for evaluating the quality of fruits and vegetables. The equipment now available is not feasible for routine quality testing; however, costs and capabilities are rapidly improving. Each sensor method is based

on the measurement of a given constituent or property; therefore its ability to measure overall quality is only as good as the relationship of that constituent or property to quality as dened for a particular purpose. Improved statistical methods for combining the inputs from several measurements into classication algorithms are being developed.

References
Abbott, J.A., 1996. Quality measurement by delayed light emission and uorescence. In: Dull, G.G., Iwamoto, M., Kawano, S. (Eds.), Nondestructive Quality Evaluation of Horticultural Crops: Proceedings International Symposium on Nondestructive Quality Evaluation of Horticultural Crops, 24th International Horticulture Congress, 26 August 1994, Kyoto, Japan. Saiwai Shobou Publishing Co, Tokyo, pp. 24 33. Abbott, J.A., Massie, D.R., 1985. Delayed light emission for early detection of chilling in cucumber and bell pepper fruit. J. Am. Soc. Hortic. Sci. 110, 42 47. Abbott, J.A., Massie, D.R., 1998. Nondestructive sonic measurement of kiwifruit rmness. J. Am. Soc. Hortic. Sci. 123, 317 322. Abbott, J.A., Childers, N.F., Bachman, G.S., Fitzgerald, J.V., Matusik, F.J., 1968. Acoustic vibration for detecting textural quality of apples. Proc. Am. Soc. Hortic. Sci. 93, 725 737. Abbott, J.A., Miller, A.R., Campbell, T.A., 1991. Detection of mechanical injury and physiological breakdown of cucumbers using delayed light emission. J. Am. Soc. Hortic. Sci. 116, 52 57. Abbott, J.A., Campbell, T.A., Massie, D.R., 1994. Delayed light emission and uorescence responses of plants to chilling. Remote Sensing Environ. 47, 87 97. Abbott, J.A., Massie, D.R., Upchurch, B.L., Hruschka, W.R., 1995. Nondestructive sonic rmness measurement of apples. Trans. ASAE 38 (5), 1461 1466. Abbott, J.A., Lu, R., Upchurch, B.L., Stroshine, R.L., 1997. Technologies for nondestructive quality evaluation of fruits and vegetables. Hortic. Rev. 20, 1 120. Akimoto, K., 1984. A method for non-destructively grading fruits and other foodstuffs. UK patent application GB2135059 A. Akimoto, K., McClure, W.F., Shimizu, K., 1995. Non-destructive evaluation of vegetable and fruit quality by visible light and MRI. In: Proceedings Automation and Robotics in Bioproduction and Processing, 3-6 November 1995, Kobe. Japan 1, 117 124. Aneshansley, D.J., Throop, J.A., Upchurch, B.L., 1997. Reectance spectra of surface defects on apples. Sensors for Nondestructive Testing. Proceedings Sensors for Nondestructive Testing International Conference, Orlando, FL,

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225 18 21 February 1997. NRAES (Northeast Reg Agric Eng Svc) Coop Extn, Ithaca, NY, pp. 143160. Armstrong, P.R., Zapp, H.R., Brown, G.K., 1990. Impulsive excitation of acoustic vibrations in apples for rmness determination. Trans. Am. Soc. Agric. Eng. 33, 1353 1359. Bajema, R.W., Hyde, G.M., Baritelle, A.L., 1998. Temperature and strain rate effects on the dynamic failure properties of potato tuber tissue. Trans. Am. Soc. Agric. Eng. 41, 733 740. Beaudry, R.M., Mir, N., Song, J., Armstrong, P., Deng, W., Timm, E., 1997. Chlorophyll uorescence: a nondestructive tool for quality measurements of stored apple fruit. Sensors for Nondestructive Testing. Proceedings Sensors for Nondestructive Testing International Conference, Orlando, FL, 18 21 February 1997. Ithaca, NY: Northeast Reg. Agric. Eng. Svc., Coop. Extn, pp. 5666. Benady, M., Simon, J.E., Charles, D.J., Miles, G.E., 1995. Fruit ripeness determination by electronic sensing of aromatic volatiles. Trans. Am. Soc. Agric. Eng. 38, 251257. Birth, G.S., 1976. How light interacts with foods. In: Gaffney, J.J. Jr. (Ed.), Quality Detection in Foods. American Society for Agricultural Engineering, St. Joseph, MI, pp. 611. Birth, G.S., Dull, G.G., Renfroe, W.T., Kays, S.J., 1985. Nondestructive spectrophotometric determination of dry matter in onions. J. Am. Soc. Hortic. Sci. 110, 297303. Blanpied, G.D., Bramlage, W.J., Dewey, D.H., LaBelle, R.L., Massey, L.M. Jr., Mattus, G.E., Stiles, W.C., Watada, A.E., 1978. A standardized method for collecting apple pressure test data. New York Food Life Sci. Bull. 74. Bourne, M.C., 1965. Studies on punch testing of apples. Food Technol. 19, 113 115. Bourne, M.C., 1979. Fruit texturean overview of trends and problems. J. Texture Stud. 10, 8394. Bourne, M.C., 1982. Food Texture and Viscosity: Concept and Measurement. Academic Press, New York. Brecht, J.K., Shewfelt, R.L., Garner, J.C., Tollner, E.W., 1991. Using x-ray computed tomography (x-ray CT) to nondestructively determine maturity of green tomatoes. HortScience 26, 45 47. Brown, G.K., Sarig, Y. Jr. (Eds.), 1994. Nondestructive Technologies for Quality Evaluation of Fruits and Vegetables: Proceedings International Workshop, US-Israel BARD Fund, Spokane, WA, 1519 June 1993. American Society for Agricultural Engineering, St. Joseph, MI. Butler, W.L., Norris, K.H., 1963. Lifetime of the long-wavelength chlorophyll uorescence. Biochim. Biophys. Acta 66, 72 77. Chan, H.T. Jr., Forbus, W.B. Jr., 1988. Delayed light emission as a biochemical indicator of papaya heat treatment. J. Food Sci. 53, 1490 1492. Chen, P., 1996. Quality evaluation technology for agricultural products. Proceedings International Conference Agricultural Machinery Engineering, Seoul, Korea, vol. 1, pp. 171 204. Chen, P., Sun, Z., 1991. A review of nondestructive methods for quality evaluation and sorting of agricultural products. J. Agric. Eng. Res. 49, 8598.

223

Chen, P., McCarthy, M.J., Kauten, R., 1989. NMR for internal quality evaluation of fruits and vegetables. Trans. Am. Soc. Agric. Eng. 32, 1747 1753. Chen, Y.R., Delwiche, S.R., Hruschka, W.R., 1995. Classication of hard red wheat by feedforward backpropagation neural networks. Cereal Chem. 72, 317 319. Chen, P., McCarthy, M.J., Kim, S.M., Zion, B., 1996. Development of a high-speed NMR technique for sensing maturity of avocados. Trans. Am. Soc. Agric. Eng. 39, 2205 2209. Cho, S.I., Stroshine, R.L., Baianu, I.C., Krutz, G.W., 1993. Nondestructive sugar content measurements of intact fruit using spin-spin relaxation times (T2) measurements by pulsed 1H magnetic resonance. Trans. Am. Soc. Agric. Eng. 36, 1217 1221. Chuma, Y., Nakaji, K., 1979. Delayed light emission as a means of color sorting of plant products. In: Linko, P., Malkki, Y., Olkku, J. (Eds.), Food Process Engineering, vol. I, pp. 314 319. Chuma, Y., Nakaji, K., Ohura, M., 1980. Maturity evaluation of bananas by delayed light emission. Trans. Am. Soc. Agric. Eng. 23 (4), 1043 1047. Chuma, Y., Nakaji, K., Tagawa, A., 1982. Delayed light emission as a means of automatic sorting of tomatoes. J. Fac. Agr. Kyushu Univ. Fukuoka 26 (4), 221 234. Clydesdale, F.M., 1978. Colorimetry methodology and applications. Crit. Rev. Food Sci. Nutr. 10, 243 301. Clark, C.J., Hockings, P.D., Joyce, D.C., Mazucco, R.A., 1997. Application of magnetic resonance imaging to preand post-harvest studies of fruit and vegetables. Postharvest Biol. Technol. 11, 1 21. DeEll, J.R., Prange, R.K., Murr, D.P., 1996. Chlorophyll uorescence as a potential indicator of controlled atmosphere disorders in Marshall McIntosh apples. HortScience 30, 1084 1085. Delwiche, M.J., Sarig, Y., 1991. A probe impact sensor for fruit rmness measurement. Trans. Am. Soc. Agric. Eng. 34, 187 192. Delwiche, M.J., Tang, S., Mehlschau, J.J., 1989. An impact force response fruit rmness sorter. Trans. Am. Soc. Agric. Eng. 32, 321 326. Dull, G.G., 1984. Measurement of cantaloupe quality with near infrared spectrophotometry. Am. Soc. Agric. Eng. Paper 84-3557. Dull, G.G., Birth, G.S., Lefer, R.G., 1989. Use of near infrared analysis for the nondestructive measurement of dry matter in potatoes. Am. Potato J. 66, 215 225. Dull, G.G., Iwamoto, M., Kawano, S. Jr. (Eds.), 1996. Nondestructive Quality Evaluation of Horticultural Crops. Proceedings International Symposium Nondestructive Quality Evaluation of Horticultural Crops, International Horticulture Congress, Kyoto, Japan, 26 August 1994. Saiwai Shobou Publishing Co, Tokyo. Faust, M., Wang, P.C., Maas, J., 1997. The use of magnetic resonance imaging in plant science. Hortic. Rev. 20, 225 266.

224

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225 Lee, K.J., Hruschka, W.R., Abbott, J.A., Noh, S.H., Chen, Y.R., 1997. Comparison of statistical methods for calibrating soluble solid measurement from NIR interactance in apples. Proc. Korean Soc. Agric. Machinery. 97 Winter Conf. 2 (1), 206 212. Lenker, D.H., Adrian, P.A., 1971. Use of x-ray for selecting mature lettuce heads. Trans. Am. Soc. Agric. Eng. 14, 894 898. Lu, R., Abbott, J.A., 1997. Finite element modeling of transient responses of apples to impulse excitation. Trans. Am. Soc. Agric. Eng. 40, 1395 1406. Lurie, S., Ronen, R., Meier, S., 1994. Determining chilling injury induction in green peppers using nondestructive pulse amplitude modulated (PAM) uorometry. J. Am. Soc. Hortic. Sci. 119, 59 62. MacFall, J.S., Johnson, G.A., 1994. The architecture of plant vasculature and transport as seen with magnetic resonance microscopy. Can. J. Bot. 72, 1561 1573. Magness, J.R. and Taylor, G.F., 1925. An improved type of pressure tester for the determination of fruit maturity. USDA Circular 350. McGlone, V.A., Schaare, P.N., 1993. The application of impact response analysis in the New Zealand fruit industry. Am. Soc. Agric. Eng. Paper 93-6537. Melcarek, P.K., Brown, G.N., 1977. The effects of chilling stress on the chlorophyll uorescence of leaves. Plant Cell Physiol. 18, 1099 1107. Minolta, 1994. Precise Color Communication. Minolta Co, Ramsey, NJ. Mir, N., Wendorf, M., Perez, R., Beaudry, R.M., 1998. Chlorophyll uorescence in relation to supercial scald development in apple. J. Am. Soc. Hortic. Sci. 123, 887 892. Mohsenin, N.N., 1977. Characterization and failure in solid foods with particular reference to fruits and vegetables. J. Texture Stud. 8, 169 193. Murakami, M., Himoto, J., Itoh, K., 1994. Analysis of apple quality by near infrared reectance spectroscopy. J. Fac. Agr. Hokkaido Univ. 66 (1), 51 61. Nakaji, K., Sugiura, Y., Chuma, Y., 1978. Delayed light emission of green tea as a means of quality evaluation. 1. Delayed light emission of fresh tea leaves. J. Soc. Agric. Mach. Jpn 39 (4), 483-490 In Japanese, with English summary. Neufeldt, V.E. (Ed.), 1988. Websters New World Dictionary, 3rd college ed. Simon and Schuster, New York. NRAES, 1997. Sensors for Nondestructive Testing. Proceedings Sensors for Nondestructive Testing International Conference, Orlando, FL, 18 21 February 1997. Northeast Reg. Agric. Eng. Svc., Coop. Extn, Ithaca, NY. Nylund, R.E., Lutz, J.M., 1950. Separation of hollow heart potato tubers by means of size grading, specic gravity, and x-ray examination. Am. Potato J. 27, 214 222. Ozer, N., Engel, B., Simon, J., 1995. Fusion classication techniques for fruit quality. Trans. Am. Soc. Agric. Eng. 38, 1927 1934. Patel, N., McGlone, V.A., Schaare, P.N., Hall, H., 1993. BerryBounce: a technique for the rapid and nondestructive measurement of rmness in small fruit. Int. Soc. Hortic. Sci. 352, 189 198.

Forbus, W.R. Jr., Senter, S.D., Chan, H.T. Jr., 1987. Measurement of papaya maturity by delayed light emission. J. Food Sci. 52, 356 360. Forbus, W.R. Jr., Dull, G.G., Smittle, D.A., 1992. Nondestructive measurement of canary melon maturity by delayed light emission. J. Food Qual. 15, 119127. Francis, F.J., 1980. Color quality evaluation of horticultural crops. HortScience 15, 1415. Galili, N., Mizrach, A., Rosenhouse, G., 1993. Ultrasonic testing of whole fruit for nondestructive quality evaluation. Am. Soc. Agric. Eng. Paper 93-6026. Gunasekaran, S., Paulsen, M.R., Shove, G.C., 1985. Optical methods for nondestructive quality evaluation of agricultural and biological materials. J. Agric. Eng. Res. 32, 209241. Haller, M.H., 1941. Fruit pressure testers and their practical applications. USDA Circular 627. Hamson, A.R., 1952. Measuring rmness of tomatoes in a breeding program. Proc. Am. Soc. Hortic. Sci. 60, 425433. Harker, F.R., Redgwell, R.J., Hallett, I.C., Murray, S.H., Carter, G., 1997. Texture of fresh fruit. Hortic. Rev. 20, 121 224. Heinemann, P.H., Varghese, Z.A., Morrow, C.T., Sommer, H.J., Crassweller, R.M., 1995. Machine vision inspection of Golden Delicious apples. Appl. Eng. Agric. 11, 901906. Hinshaw, W.S., Bottomley, P.A., Holland, G.N., 1979. A demonstration of the resolution of NMR imaging in biological systems. Experientia 35, 12681269. Hruschka, W.R., 1987. Data analysis: wavelength selection methods. In: Williams, P., Norris, K. (Eds.), Near-Infrared Technology in the Agricultural and Food Industries. American Association of Cereal Chemists, St. Paul, MN, pp. 35 55. Hunter, R.S., Harold, R.W., 1987. The Measurement of Appearance. Wiley-Interscience, New York. Jackman, R.L., Marangoni, A.G., Stanley, D.W., 1990. Measurement of tomato fruit rmness. HortScience 25, 781783. Jacob, R.C., Romani, R.J. and Sprock, C.M., 1965. Fruit sorting by delayed light emission. Trans. Am. Soc. Agric. Eng. 8, 18 19, 24. Katayama, K., Komaki, K., Tamiya, S., 1996. Prediction of starch, moisture, and sugar in sweetpotato by near infrared transmittance. HortScience 31, 10031006. Kawano, S., Watanabe, H., Iwamoto, M., 1992. Determination of sugar content in intact peaches by near infrared spectroscopy with ber optics in interactance mode. J. Jpn. Soc. Hortic. Sci. 61, 445 451. Kawano, S., Fujiwara, K., Iwamoto, M., 1993. Nondestructive determination of sugar content in Satsuma mandarin using near infrared (NIR) transmittance. J. Jpn. Soc. Hortic. Sci. 62, 465 470. Keagy, P.M., Parvin, B., Schatzki, T.F., 1996. Machine recognition of navel orange worm damage in x-ray images of pistachio nuts. Lebensm.-Wiss. Technol. 29, 140145. Lancaster, J.E., Lister, C.E., Reay, P.F., Triggs, C.M., 1997. Inuence of pigment composition on skin color in a wide range of fruits and vegetables. J. Am. Soc. Hortic. Sci. 122, 594 598.

J.A. Abbott / Posthar6est Biology and Technology 15 (1999) 207225 Peleg, K., Ben-Hanan, U., Hinga, S., 1990. Classication of avocado by rmness and maturity. J. Texture Stud. 21, 123 129. Pieris, K.H.S., Dull, G.G., Lefer, R.G., Kays, S.J., 1998. Near-infrared sepctrometric method for nondestructive determination of soluble solids contnet of peaches. J. Am. Soc. Hortic. Sci. 123, 898 905. Prussia, S.E., Astleford, J.J., Hewlett, B., Hung, Y.C., 1994. Non-destructive rmness measuring device. US Patent 5,372,030. Saltveit, M. Jr., 1991. Determining tomato fruit maturity with nondestructive in vivo nuclear magnetic resonance imaging. Postharvest Biol. Technol. 1, 153159. Saltveit, M.E., Upadhyaya, S.K., Happ, J.F., Caveletto, R., OBrien, M., 1985. Maturity determination of tomatoes using acoustic methods. Am. Soc. Agric. Eng. Paper 853536. Schatzki, T.F., Witt, S.C., Wilkins, D.E., Lenker, D.H., 1981. Characterization of growing lettuce from density contours I. Head selection. Pattern Recogn. 13, 333340. Schatzki, T.F., Haff, R.P., Young, R., Can, I., Le, L.C., Toyofuku, N., 1997. Defect detection in apples by means of x-ray imaging. Sensors for Nondestructive Testing: Proceedings Sensors for Nondestructive Testing International Conference, Orlando, FL, 1821 February 1997. NRAES (Northeast Reg Agric Eng Svc) Coop Extn, Ithaca, NY, pp. 161 171. Schreiber, U., Groberman, L., Vidaver, W., 1975. Portable, solid-state uorometer for the measurement of chlorophyll uorescence induction in plants. Rev. Sci. Instrum. 46, 538 542. Seiden, P., Bro, R., Poll, L., Munck, L., 1996. Exploring uorescence spectra of apple juice and their connection to quality parameters by chemometrics. J. Agric. Food Chem. 44, 3202 3205. Shewfelt, R.L., 1999. What is quality? Postharvest Biol. Technol. (in press). Shmulevich, I., Galili, N., Benichou, N., 1995. Development of a nondestructive method for measuring the shelf-life of mango fruit. Proceedings Food Processing Automation IV Conference, Chicago, IL, 35 November. American Society for Agricultural Engineering, St. Joseph, MI, pp. 275287. Simon, J.E., Hetzroni, A., Bordelon, Miles, G.E., and Charles, D.J., 1996. Electronic sensing of aromatic volatiles for quality sorting of blueberry fruit. J. Food Sci. 61, 967969, 972. Slaughter, D.C., Barrett, D., Boersig, M., 1996. Nondestructive determination of soluble solids in tomatoes using near infrared spectroscopy. J. Food Sci. 61, 695697. Smillie, R.M., Nott, R., 1979. Assay of chilling injury in wild and domestic tomatoes based on photosystem activity of the chilled leaves. Plant Physiol. 63, 796801. Smillie, R.M., Hetherington, S.E., Nott, R., Chaplin, G., Wade, N.L., 1987. Applications of chlorophyll uorescence to the postharvest physiology and storage of mango and banana fruit and the chilling tolerance of mango cultivars. ASEAN Food J. 3, 55 59. Song, H., Delwiche, S.R., Chen, Y.R., 1995. Neural network

225

classication of wheat using single kernel near-infrared transmittance spectra. Opt. Eng. 34, 2927 2934. Stone, M.L., Zhang, X., Brusewitz, G.H., Chen, D.D., 1994. Watermelon maturity determination in the eld with acoustic impedance techniques. Am. Soc. Agric. Eng. Paper 94-024. Stroshine, R.L., Wai, W.K., Keener, K.M., Krutz, G.W., 1994. New developments in fruit ripeness sensing using magnetic resonance. Am. Soc. Agric. Eng. Paper 94-6539. Sugiyama, J., Otobe, K., Kikuchi, Y., 1994. A novel rmness meter for fruits and vegetables. Am. Soc. Agr. Eng. Paper 94-6030. Throop, J.A., Aneshansley, D.J., 1997. Apple damage segmentation utilizing reectance spectra of the defect. Am. Soc. Agr. Eng. Paper 97-3078. Tian, M.S., Woolf, A.B., Bowen, J.H., Ferguson, I.B., 1996. Changes in color and chlorophyll uorescence of broccoli orets following hot water treatment. J. Am. Soc. Hortic. Sci. 121, 310 313. Toivonen, P.M.A., 1992. Chlorophyll uorescence as a nondestructive indicator of freshness in harvested broccoli. HortScience 27, 1014 1015. Tollner, E.W., Hung, Y.C., Upchurch, B.L., Prussia, S.E., 1992. Relating x-ray absorption to density and water content in apples. Trans. Am. Soc. Agric. Eng. 35, 1921 1928. Tollner, E.W., Brecht, J.K., Upchurch, B.L., 1993. Nondestructive evaluation: detection of external and internal attributes frequently associated with quality or damage. In: Prussia, S.E., Shewfelt, R.L. (Eds.), Postharvest Handling: A Systems Approach. Academic Press, New York, pp. 225 255. Tollner, E.W., Hung, Y.C., Maw, B.W., Sumner, D.R., Gitaitis, R.D., 1995. Nondestructive testing for identifying poor quality onions. Proc. Soc. Photo-Opt. Instrum. Eng. 2345, 392 397. Uozumi, J. L, Kawano, S., Iwamoto, M., Nishinari, K., 1987. Spectrophotometric system for quality evaluation of unevenly colored food. Nippon Shokuhin Kogyo Gakkaishi 34, 163-170 in Japanese. Upchurch, B.L., Miles, G.E., Stroshine, R.L., Furgason, E.S., Emerson, F.H., 1987. Ultrasonic measurement for detecting apple bruises. Trans. Am. Soc. Agric. Eng. 30, 803 809. Upchurch, B.L., Throop, J.A., Aneshansley, D.J., 1994. Inuence of time, bruise-type, and severity on near-infared reectance from apple surfaces for automatic bruise detection. Trans. Am. Soc. Agric. Eng. 37 (5), 1571 1575. Wang, S.Y., Wang, P.C., Faust, M., 1988. Non-destructive detection of watercore in apple with nuclear magnetic resonance imaging. Sci. Hortic. 35, 227 334. Williams, P., Norris, K. (Eds.), 1987. Near-Infrared Technology in the Agricultural and Food Industries. American Association of Cereal Chemists, St. Paul, MN. Woolf, A.B., Laing, W.A., 1996. Avocado fruit skin uorescence following hot water treatments and pretreatments. J. Am. Soc. Hortic. Sci. 121, 147 151. Zapp, H.R., Ehlert, S.H., Brown, G.K., Armstrong, P.R., Sober, S.S., 1990. Advanced instrumented sphere (IS) for impact measurements. Trans. Am. Soc. Agric. Eng. 33, 955 960.